So LM317 has about 2.5 volt dropout at full current draw, plus a volt or so dropped across the rectifier diodes, means you need at least 24 rectified volts or so from the transformer for 18 volt output. Yet the specified transformer only has a 12 volt secondary? I noticed in OPC's build there is a 15 volt secondary transformer, but even that seems like it is cutting it a bit too close to the dropout. It seems to me that part number TE2260-ND from digikey, with 18 volt secondaries would be more appropriate to ensure no dropout on the 18 volt supply rails.
One day we'll be able to take something like this and get 1000 W or more out of it
Please aim it Eastward. My dogz earz are very sensitive
Hi Guys,
Looks like there are a few questions I should address.
tommy thong:
You could parallel more buffers and indeed get more power into lower impedance loads, but if you just solder them to the existing pins sticking up in the air, then dissipation will be a serious problem. You would need to re-design the PCB for that to work properly.
You also need to be careful to dial in the available voltage swing, the load you plan to drive, and the resulting current required. There's a reason I stopped at 8 buffers, and that's because they provide all the current needed to drive an 8 ohm speaker into voltage clipping with 18V rails. Adding 8 more buffers will not result in any more power into an 8 ohm load unless you also increase the rails, which cannot go any higher with the LME49990 in the front end.
bcg27:
Good catch... I mistakenly put the 12V transformer in the BOM when it definitely needs to be a 15V transformer as I used on my boards.
After running my testing, I would actually suggest you try an 18V transformer as it will help keep the rails up under full load, and hence provide more power output. You may need to use higher voltage caps before the regs though.
Another option would be to use a higher VA 15VAC transformer mounted off-board with the flying leads soldered to the correct pads. A higher VA unit wouldn't sag quite so badly at full power. I was limited in board area which is the reason I want with the 25VA unit. 50-100VA would be a better choice.
Overall though, the 15VAC 25VA unit does alright. It's designed for 115VAC so when used at 120VAC here in NA you get a little more voltage. The diodes also have a very low forward voltage at about 0.32V @ 2A. Just the same though, the rails definitely do sag below 18V at full power. I would suggest either lower voltage drop regs, or a larger VA transformer if you think that is an issue for you.
Cheers,
Owen
Looks like there are a few questions I should address.
tommy thong:
You could parallel more buffers and indeed get more power into lower impedance loads, but if you just solder them to the existing pins sticking up in the air, then dissipation will be a serious problem. You would need to re-design the PCB for that to work properly.
You also need to be careful to dial in the available voltage swing, the load you plan to drive, and the resulting current required. There's a reason I stopped at 8 buffers, and that's because they provide all the current needed to drive an 8 ohm speaker into voltage clipping with 18V rails. Adding 8 more buffers will not result in any more power into an 8 ohm load unless you also increase the rails, which cannot go any higher with the LME49990 in the front end.
bcg27:
Good catch... I mistakenly put the 12V transformer in the BOM when it definitely needs to be a 15V transformer as I used on my boards.
After running my testing, I would actually suggest you try an 18V transformer as it will help keep the rails up under full load, and hence provide more power output. You may need to use higher voltage caps before the regs though.
Another option would be to use a higher VA 15VAC transformer mounted off-board with the flying leads soldered to the correct pads. A higher VA unit wouldn't sag quite so badly at full power. I was limited in board area which is the reason I want with the 25VA unit. 50-100VA would be a better choice.
Overall though, the 15VAC 25VA unit does alright. It's designed for 115VAC so when used at 120VAC here in NA you get a little more voltage. The diodes also have a very low forward voltage at about 0.32V @ 2A. Just the same though, the rails definitely do sag below 18V at full power. I would suggest either lower voltage drop regs, or a larger VA transformer if you think that is an issue for you.
Cheers,
Owen
Good points. If you want to keep the nice compact everything on pcb build it seems like the best option might be qusp's with a ldo regulator, since it doesn't really look like there is room for 4700 uF 35 V caps. Although it probably won't hurt anything to use 3900 uF 35 V caps with the 18 vac transformer instead.
One more question: is the heatsink you used on the lme49600 buffers a standard part or did you machine it yourself? If it's standard would you mind sharing the part number? Thanks
One more question: is the heatsink you used on the lme49600 buffers a standard part or did you machine it yourself? If it's standard would you mind sharing the part number? Thanks
I should have included the heastsink on the BOM, but I forgot. I'll post an updated BOM and fix the transformer and add the heatsink.
It's a standard part:
Digi-Key - 102-1489-ND (Manufacturer - VHS-95)
The spacing is also standard "half brick" spacing so any heatsink designed for a half brick power supply should work.
Here's a pimpin' one:
MPC5861
What I'm still struggling to find is a decent gap filling thermal pad that doesn't cost a fortune. I had two scrap pieces of 0.2" thick high compliance stuff from work and it's pretty much perfect as it forms around the buffers and makes good contact with the PCB which is where all the heat goes.
If anyone can find a thermal pad then please post a link up here!
Cheers,
Owen
It's a standard part:
Digi-Key - 102-1489-ND (Manufacturer - VHS-95)
The spacing is also standard "half brick" spacing so any heatsink designed for a half brick power supply should work.
Here's a pimpin' one:
MPC5861
What I'm still struggling to find is a decent gap filling thermal pad that doesn't cost a fortune. I had two scrap pieces of 0.2" thick high compliance stuff from work and it's pretty much perfect as it forms around the buffers and makes good contact with the PCB which is where all the heat goes.
If anyone can find a thermal pad then please post a link up here!
Cheers,
Owen
Thanks for the link. .2 inches seems to be quite thick for thermal pads and is expensive everywhere I can find. However thinner sheets like 1 mm seem to be available with reasonable prices - stacking would work maybe?
Check out the stuff at this link:
http://www.frozencpu.com/cat/l3/g8/c487/s1290/list/p1/Thermal-Thermal_Pads_Tape-Thermal_Pad_-_10mm_-Page1.html
Check out the stuff at this link:
http://www.frozencpu.com/cat/l3/g8/c487/s1290/list/p1/Thermal-Thermal_Pads_Tape-Thermal_Pad_-_10mm_-Page1.html
I think stacking may work but would be a slight compromise. I imagine it would have a reasonable chance of air bubbles being created between layers which would, for obvious reasons, have an adverse impact.
opc I'm not familiar with specs of thermal pads but did the spares you used have a rating/specification for hardness that we should be aiming for?
Ebay, may also have some options from a quick search I found this -
100mm*100mm*5mm Silicone Heat Conductive Thermal Pads | eBay
opc I'm not familiar with specs of thermal pads but did the spares you used have a rating/specification for hardness that we should be aiming for?
Ebay, may also have some options from a quick search I found this -
100mm*100mm*5mm Silicone Heat Conductive Thermal Pads | eBay
probably there is something on the pcb thats .2" higher than the buffers
edit: nope, just checked, yes opc, why is it there is a thermal gap that needs filling? from the standoffs if you use them between pcb and heatsink?